CN114298834A - Personal credit evaluation method and system based on self-organizing mapping network - Google Patents
Personal credit evaluation method and system based on self-organizing mapping network Download PDFInfo
- Publication number
- CN114298834A CN114298834A CN202111683268.6A CN202111683268A CN114298834A CN 114298834 A CN114298834 A CN 114298834A CN 202111683268 A CN202111683268 A CN 202111683268A CN 114298834 A CN114298834 A CN 114298834A
- Authority
- CN
- China
- Prior art keywords
- personal credit
- data
- evaluation
- personal
- self
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 101
- 238000013507 mapping Methods 0.000 title claims abstract description 32
- 238000013528 artificial neural network Methods 0.000 claims abstract description 39
- 238000013210 evaluation model Methods 0.000 claims abstract description 36
- 230000009467 reduction Effects 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims description 57
- 230000006870 function Effects 0.000 claims description 44
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000000513 principal component analysis Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 21
- 239000013598 vector Substances 0.000 claims description 11
- 230000014509 gene expression Effects 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 8
- 230000003044 adaptive effect Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 238000011946 reduction process Methods 0.000 claims description 4
- 239000003795 chemical substances by application Substances 0.000 claims description 3
- 230000003862 health status Effects 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 claims description 2
- 238000012847 principal component analysis method Methods 0.000 abstract description 5
- 210000004556 brain Anatomy 0.000 abstract description 2
- 238000003909 pattern recognition Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000036961 partial effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Landscapes
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a personal credit evaluation method based on a self-organizing mapping network, which comprises the following steps: constructing a personal credit index system; collecting personal credit data; reducing dimension of the features; constructing a self-organizing mapping neural network, generating a personal credit evaluation model, importing the personal credit data sample subjected to dimensionality reduction into the personal credit evaluation model, and training the personal credit evaluation model; personal credit assessment; and outputting an evaluation result. The invention can compress the personal credit attribute space by adopting a principal component analysis method, thereby reducing the complexity of personal credit evaluation; on the basis of the principal component variable generated by dimension reduction, the self-organizing mapping neural network simulating the self-organizing characteristic of the human brain is used for carrying out pattern recognition on the input personal credit sample, so that the accurate evaluation on the personal credit category is realized.
Description
Technical Field
The invention relates to the technical field of credit assessment, in particular to a personal credit assessment method and system based on a self-organizing mapping network.
Background
With the rapid development of socioeconomic of China, the living standard of people is gradually improved, the income in the current period cannot meet the consumption demand of people, and the consumption credit market is increased unprecedentedly. To circumvent the risk of credit consumption, commercial banks need to make personal credit assessments of credit consumption applicants and a comprehensive assessment of their potential for repayment on time to make a decision on whether to loan or not. The personal credit assessment is the qualitative analysis of the credit condition of a credit applicant based on the personal experience and knowledge of a credit assessment expert at the earliest time, the method depends on the professional experience of the credit assessment expert, and the method has strong subjectivity and cannot comprehensively and objectively evaluate the credit condition of the credit applicant. With the development of computer computing power and the rapid development of mathematical statistics and artificial intelligence algorithms, a large number of statistical methods and group intelligence algorithms are applied to the field of credit evaluation.
In many fields of research and applications, the spatial dimension of object attributes is generally high in order to more fully study the relevant properties of the objects involved. In most cases, there is a correlation between object attribute parameters, and there must be a dominant common factor between these object attributes, so it is necessary to compress the attribute space, to prevent the information provided by the sample data from cross-overlapping, and possibly even to mask the true features of the object under study, thereby increasing the complexity of the analysis problem. Therefore, people hope that the selected explanatory variables can fully reflect the overall characteristics of the problem to be researched, and hope that the number of the explanatory variables is as small as possible, principal component analysis is a method for solving the problem, the principal component analysis is a linear combination of the original explanatory variables with certain correlation, and the original variables are replaced by the combined comprehensive variables, so that the number of the explanatory variables is reduced, and the representativeness of the selected index is ensured.
2020.12.04, the patent number CN112037012A discloses an internet financial credit evaluation method based on a PSO-BP neural network, which includes obtaining the true value of the result of information, normalizing the obtained data, analyzing and reducing the dimensions of the principal components, dividing a test set and a training set, initializing the number of input nodes, the number of output nodes and the number of hidden layer nodes of the BP neural network, continuously adjusting the weight and the threshold of the network by using the traditional gradient descent method and back propagation to construct a BP neural network model, optimizing the connection weight and the threshold by using a particle swarm algorithm to obtain the PSO-BP neural network model, wherein the verification set is used for verification and optimization, and the model is deployed to an application system to extract the characteristic parameters of the data of a real-time application client and predict and classify the data. The invention greatly improves the convergence rate of the BP neural network, and the obtained credit evaluation model of the PSO-BP neural network can accurately and quickly realize the credit evaluation on an internet financial applicant, effectively improves the service timeliness of application approval, and reduces the wind control cost and the application fraud risk.
2016.12.07, the invention with patent number CN106204246A discloses a BP neural network credit evaluation method based on principal component analysis, which comprises the steps of combing government data related to individuals from bank data, combining the credit evaluation result of the bank to the individuals, forming sample data, improving the prediction performance after normalization processing of the sample data, reducing the dimension of the sample data by using the principal component analysis method, solving the complex index and multidimensional data types, meeting the requirement of big data processing, and using the credit evaluation result of the bank to the individuals as the reference for training the BP neural network model, thereby constructing a credit evaluation model based on government big data, overcoming the subjectivity of expert scoring, providing credit inquiry for enterprises or individuals, supplementing the credit system of financial institutions, having higher classification accuracy and practicability, and better evaluation effect.
However, both of the above two inventions use a BP neural network model, and a set of training sets is fed into the network using a supervised learning algorithm, and the connection weights are adjusted according to the difference between the actual output and the expected output of the network. But in real credit assessment applications, sometimes giving mentor information has certain difficulties. In addition, the credit indicator settings of the two inventions are relatively fixed, the applicable scene is single, the existing relationship with the application object (such as a commercial bank) is not involved in the selected credit indicator, and the accuracy of the actual evaluation result is not high.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a personal credit assessment method and a personal credit assessment system based on a self-organizing mapping network, which can construct a personal credit index system according to the purpose and the requirement of personal credit assessment, and then compress a personal credit attribute space by adopting a principal component analysis method to reduce the complexity of personal credit assessment; on the basis of the principal component variable generated by dimension reduction, the self-organizing mapping neural network simulating the self-organizing characteristic of the human brain is used for carrying out clustering pattern recognition on the input personal credit sample, so that the accurate evaluation on the personal credit category is realized.
In order to achieve the purpose, the invention adopts the following technical scheme:
a personal credit assessment method based on an ad hoc mapping network, the personal credit assessment method comprising the steps of:
s1, constructing a personal credit index system: according to the purpose and the requirement of personal credit evaluation, acquiring an index which can most reflect the personal credit characteristics of an evaluation object, and constructing and obtaining a personal credit index system; the personal credit evaluation index can comprise contents such as personal natural characteristics, professional characteristics and bank relation;
s2, collecting personal credit data: acquiring personal credit data of a plurality of consumption credit applicants from different data sources according to a personal credit index system to generate a personal credit data sample;
s3, feature dimension reduction: performing principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of an evaluation object, and obtaining a personal credit data sample after dimension reduction;
s4, constructing a self-organizing mapping neural network, generating a personal credit evaluation model, importing the personal credit data sample after dimensionality reduction into the personal credit evaluation model, and training the personal credit evaluation model; the personal credit evaluation model comprises a processing unit array, a comparison selection unit, a local interconnection unit and an adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptive unit modifies the parameters of the excited processing unit to increase its output value corresponding to the specific input discriminant function;
s5, personal credit assessment: collecting personal credit data of an object to be evaluated, replacing original variables with combined comprehensive variables according to the principal component analysis result of the personal credit data, and inputting the combined comprehensive variables into a self-organizing mapping neural network for personal credit evaluation;
and S6, outputting an evaluation result: and outputting the personal credit evaluation result of the self-organizing mapping neural network to a commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, in step S1, the index that can best reflect the personal credit characteristics of the evaluation target may include age, gender, marital status, health status, culture degree, unit property, industry status, post age, income per family, presence or absence of employees in the bank, account in the bank, business transaction, and borrowing status.
Further, in step S3, the personal credit data is subjected to principal component analysis using IBM SPSS Statistics 22 tool.
Further, in step S3, the feature dimension reduction process includes the following steps:
s31, feature centering: standardizing the imported personal credit data to obtain a standardized data matrix;
s32, calculating a covariance matrix: establishing a corresponding covariance matrix according to the standardized data matrix, wherein the covariance matrix is used for reflecting statistical indexes of the degree of closeness of correlation among the standardized data, and the larger the value is, the more necessary principal component analysis is carried out on the data;
s33, calculating characteristic values and characteristic vectors: solving the eigenvalue, the principal component contribution rate and the accumulated variance contribution rate according to the covariance matrix, and determining the number of the principal components;
s34, selecting a characteristic vector corresponding to the large characteristic value: sorting the eigenvalues in a descending order, and selecting eigenvectors corresponding to the largest eigenvalues as column vectors to form an eigenvector matrix;
s35, establishing an initial factor load matrix, and explaining principal components: the factor load is a correlation coefficient R (Zi, Xi) of the main component Zi and the original index Xi, and is used for expressing the degree of correlation between the main component Zi and each financial ratio;
and S36, solving the function expression of each principal component, and obtaining the personal credit data of the consumption credit applicant after dimensionality reduction according to the function expression of each principal component and the personal credit data of the consumption credit applicant acquired from a plurality of data sources.
Further, in step S4, the construction process of the personal credit evaluation model includes the following sub-steps:
s41, creating a self-organizing map neural network by using a MATLAB R2016b tool, and initializing the network; the default value of the structure function is 'hextop', the default value of the distance function is 'linkdist', the default value of the learning rate of the classification stage is 0.9, the default value of the learning step length of the classification stage is 1000, the default value of the learning rate of the tuning stage is 0.02, and the default value of the neighborhood distance of the tuning stage is 1;
s42, setting the competition layer to be a 6 multiplied by 6 hexagonal structure and training times to be 300;
s43, inputting personal credit data samples after dimension reduction;
s44, searching network winning nodes;
s45, defining a winning neighborhood;
s46, adjusting the network weight;
s47, inputting new sample data and repeating the above learning process until the learning rate decays to 0 or a predetermined positive fraction.
Based on the foregoing method, the present invention also provides a personal credit evaluation system based on self-organizing map network, the personal credit evaluation system comprising:
the personal credit index construction unit is used for acquiring indexes which can best reflect the personal credit characteristics of an evaluation object according to the purpose and the requirement of personal credit evaluation, and constructing and obtaining a personal credit index system; the personal credit evaluation index can comprise contents such as personal natural characteristics, professional characteristics and bank relation;
the personal credit data acquisition unit is used for acquiring personal credit data of the consumption credit applicant from different data sources through a personal credit data acquisition agent according to a personal credit index system;
the personal credit data dimension reduction unit is used for carrying out principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of an evaluation object and obtaining the personal credit data after dimension reduction;
the personal credit evaluation model is constructed on the basis of a self-organizing mapping neural network and comprises a processing unit array, a comparison selection unit, a local interconnection unit and a self-adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptive unit modifies the parameters of the excited processing unit to increase its output value corresponding to the specific input discriminant function;
the model training unit is used for repeatedly inputting all personal credit data into the self-organizing mapping neural network for training to enable the weight value to tend to be stable, and the trained self-organizing mapping neural network is generated to serve as a personal credit evaluation model;
the personal credit evaluation unit is used for calling the personal credit data dimension reduction unit to perform principal component analysis results on the acquired personal credit data of the object to be evaluated, replacing original variables with combined comprehensive variables, and inputting the combined comprehensive variables into a personal credit evaluation model to perform personal credit evaluation;
and the personal credit evaluation result output unit is used for outputting the personal credit evaluation result output by the personal credit evaluation model to the commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
Further, the personal credit evaluation system also comprises a personal credit database unit which is used for storing personal credit evaluation results, and an authorization mechanism is responsible for daily data maintenance and management.
The invention has the beneficial effects that:
firstly, the personal credit assessment method and system based on the self-organizing mapping network can construct a personal credit index system according to the purpose and the requirement of personal credit assessment, and on the basis, the dimensionality of personal credit data is reduced by using a principal component analysis method, so that the noise and redundant data of an independent variable data set can be reduced, the calculation amount of subsequent personal credit assessment is reduced, and the accuracy of a personal credit assessment model is improved.
Secondly, the self-organizing map neural network adopted by the personal credit evaluation method and the self-organizing map neural network based on the self-organizing map network is a network structure without instructor, compared with the network structure with instructor, the self-organizing map neural network has better vector quantization function, fast clustering capability and information fusion capability, not only can find unknown clusters in personal credit data, but also can keep the topological structures of the clusters and the personal credit data.
Drawings
Fig. 1 is a flow chart of the personal credit assessment method based on the self-organizing map network of the invention.
FIG. 2 is a schematic diagram of the dimension reduction process of the present invention.
FIG. 3 is a schematic diagram of the training process of the self-organizing map network of the present invention.
Fig. 4 is a schematic diagram of the personal credit evaluation system structure based on the self-organizing map network.
Fig. 5 is a schematic structural diagram of the self-organizing map network of the present invention.
FIG. 6 is a schematic diagram of the distance situation directly adjacent to neurons of the present invention.
FIG. 7 is a schematic diagram of the classification of each neuron according to the present invention.
FIG. 8 is a Credit applicant U1The personal credit data is used as an evaluation model to simulate the input data, and the classification condition of each neuron is shown schematically.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
It should be noted that the terms "upper", "lower", "left", "right", "front", "back", etc. used in the present invention are for clarity of description only, and are not intended to limit the scope of the present invention, and the relative relationship between the terms and the terms is not limited by the technical contents of the essential changes.
Fig. 1 is a flow chart of the personal credit assessment method based on the self-organizing map network of the invention. Referring to fig. 1, the personal credit evaluation method includes the steps of:
s1, constructing a personal credit index system: according to the purpose and the requirement of personal credit evaluation, acquiring an index which can most reflect the personal credit characteristics of an evaluation object, and constructing and obtaining a personal credit index system; the personal credit evaluation index comprises personal natural characteristics, professional characteristics and bank relation. Preferably, the most obvious indicators of the personal credit characteristics of the evaluation object comprise age, gender, marital status, health status, culture degree, unit property, industry status, position age, family income, whether the employee is in the current bank, the account in the current bank, business transaction and borrowing.
S2, collecting personal credit data: according to the personal credit index system, personal credit data of a plurality of consumption credit applicants are acquired from different data sources, and personal credit data samples are generated.
S3, feature dimension reduction: and carrying out principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of the evaluation object, and obtaining a personal credit data sample after dimension reduction.
Illustratively, the IBM SPSS Statistics 22 tool was used to perform principal component analysis on personal credit data. Referring to fig. 2, the dimension reduction process includes the following sub-steps:
s31, feature centering: and carrying out standardization processing on the imported personal credit data to obtain a standardized data matrix.
S32, calculating a covariance matrix: and establishing a corresponding covariance matrix according to the standardized data matrix, wherein the covariance matrix is used for reflecting a statistical index of the degree of closeness of correlation among the standardized data, and the larger the value is, the more necessary the principal component analysis is carried out on the data.
S33, calculating characteristic values and characteristic vectors: and solving the eigenvalue, the principal component contribution rate and the accumulated variance contribution rate according to the covariance matrix, and determining the number of the principal components.
S34, selecting a characteristic vector corresponding to the large characteristic value: sorting the eigenvalues in the descending order, and selecting the eigenvectors corresponding to the largest eigenvalues as column vectors to form an eigenvector matrix.
S35, establishing an initial factor load matrix, and explaining principal components: the factor load is a correlation coefficient R (Zi, Xi) of the principal component Zi to the original index Xi, and is used to express the degree of correlation between the principal component and each financial ratio.
And S36, solving the function expression of each principal component, and obtaining the personal credit data of the consumption credit applicant after dimensionality reduction according to the function expression of each principal component and the personal credit data of the consumption credit applicant acquired from a plurality of data sources.
S4, constructing a self-organizing mapping neural network, generating a personal credit evaluation model, importing the personal credit data sample after dimensionality reduction into the personal credit evaluation model, and training the personal credit evaluation model; the personal credit evaluation model comprises a processing unit array, a comparison selection unit, a local interconnection unit and an adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptation unit modifies the parameters of the activated processing unit to increase its output value corresponding to the particular input discriminant function.
Illustratively, referring to fig. 3, in step S4, the construction process of the personal credit assessment model includes the following sub-steps:
s41, creating a self-organizing map neural network by using a MATLAB R2016b tool, and initializing the network; the default value of the structure function is 'hextop', the default value of the distance function is 'linkdist', the default value of the learning rate of the classification stage is 0.9, the default value of the learning step length of the classification stage is 1000, the default value of the learning rate of the tuning stage is 0.02, and the default value of the neighborhood distance of the tuning stage is 1.
S42, setting the competition layer to be a 6 multiplied by 6 hexagonal structure and training times to be 300.
And S43, inputting the personal credit data sample after dimension reduction.
And S44, searching a network winning node.
S45, defining a winning neighborhood.
And S46, adjusting the network weight.
S47, inputting new sample data and repeating the above learning process until the learning rate decays to 0 or a predetermined positive fraction.
S5, personal credit assessment: and collecting personal credit data of the object to be evaluated, replacing original variables with combined comprehensive variables according to the principal component analysis result of the personal credit data, and inputting the combined comprehensive variables into a self-organizing mapping neural network for personal credit evaluation.
And S6, outputting an evaluation result: and outputting the personal credit evaluation result of the self-organizing mapping neural network to a commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
The invention constructs an evaluation index system from three aspects: on the basis of personal natural characteristics, professional characteristics and bank relation, the dimensionality of personal credit data is reduced by using a principal component analysis method, noise and redundant data of a self-variable data set can be reduced, the calculation amount of subsequent personal credit evaluation is reduced, and the accuracy of a personal credit evaluation model is improved. The adopted self-organizing mapping neural network is an unsupervised neural network, is different from the reverse transfer training of a general neural network based on a loss function, applies a competitive learning strategy, gradually optimizes the network by depending on mutual competition among neurons, uses a neighbor relation function to maintain the topological structure of an input space, can automatically and accurately classify input modes, has good vector quantization function, fast clustering capability and information fusion capability, can find unknown clusters in personal credit data, and can keep the topological structures of the clusters and the personal credit data.
Based on the foregoing method, referring to fig. 4, the present embodiment also refers to a personal credit evaluation system based on a self-organizing map network, which includes a personal credit index construction unit, a personal credit data collection unit, a personal credit data dimension reduction unit, a personal credit evaluation model, a model training unit, a personal credit evaluation unit, and a personal credit evaluation result output unit.
The personal credit index construction unit is used for acquiring indexes which can best reflect the personal credit characteristics of an evaluation object according to the purpose and the requirement of personal credit evaluation, and constructing and obtaining a personal credit index system; the construction of the credit index is the basis of credit evaluation, and the personal credit evaluation index comprises personal natural characteristics, professional characteristics and bank relation.
And the personal credit data acquisition unit is used for acquiring the personal credit data of the consumption credit applicant from different data sources through the personal credit data acquisition agent according to the personal credit index system.
And the personal credit data dimension reduction unit is used for carrying out principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of the evaluation object and obtaining the personal credit data after dimension reduction. Due to the fact that personal credit indexes are numerous, high correlation is shown among partial indexes, data redundancy is easily caused, and evaluation effect is affected.
The personal credit evaluation model is constructed on the basis of a self-organizing mapping neural network and comprises a processing unit array, a comparison selection unit, a local interconnection unit and a self-adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptation unit modifies the parameters of the activated processing unit to increase its output value corresponding to the particular input discriminant function.
And the model training unit is used for repeatedly inputting all personal credit data into the self-organizing mapping neural network for training to enable the weight value to tend to be stable, and generating the trained self-organizing mapping neural network as a personal credit evaluation model.
And the personal credit evaluation unit is used for calling the personal credit data dimension reduction unit to perform principal component analysis results on the acquired personal credit data of the object to be evaluated, replacing original variables with combined comprehensive variables, and inputting the combined comprehensive variables into the personal credit evaluation model to perform personal credit evaluation.
And the personal credit evaluation result output unit is used for outputting the personal credit evaluation result output by the personal credit evaluation model to the commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
Examples of the invention
Assuming that a certain commercial bank carries out personal credit evaluation on a consumption credit applicant, the personal credit evaluation method provided by the invention comprises the following steps:
the first step is as follows: a personal credit indicator is constructed. In this embodiment, 14 indexes shown in table 1 are used as the personal credit feature indexes.
TABLE 1 personal Credit assessment index
The second step is that: personal credit data is collected. Credit applicant personal credit data is collected from a variety of data sources, as shown in Table 2:
TABLE 29 personal Credit data of applicants
| Applicant(s) of the application | X1 | X2 | X3 | X4 | X5 | X6 | X7 | X8 | X9 | X10 | X11 | X12 | X13 | X14 |
| U1 | 6 | 2 | 5 | 5 | 8 | 6 | 4 | 6 | 3 | 9 | 2 | 6 | 4 | 5 |
| |
4 | 2 | 5 | 5 | 6 | 6 | 4 | 4 | 3 | 6 | 2 | 6 | 4 | 5 |
| |
4 | 2 | 4 | 5 | 8 | 4 | 4 | 6 | 3 | 9 | 2 | 4 | 4 | 4 |
| |
3 | 2 | 3 | 3 | 4 | 4 | 2 | 4 | 2 | 4 | 0 | 4 | 2 | 4 |
| |
4 | 1 | 2 | 3 | 2 | 3 | 2 | 4 | 2 | 4 | 0 | 4 | 2 | 4 |
| |
3 | 1 | 3 | 3 | 4 | 3 | 2 | 2 | 1 | 3 | 2 | 4 | 0 | 4 |
| |
2 | 1 | 2 | -1 | 2 | 1 | 1 | 2 | 1 | 1 | 0 | 0 | 0 | -5 |
| |
2 | 1 | 2 | 3 | 2 | 1 | 1 | 2 | 2 | 1 | 0 | 0 | 2 | 4 |
| |
3 | 1 | 3 | -1 | 1 | 2 | 2 | 2 | 1 | 3 | 2 | 0 | 0 | -5 |
The third step: principal component analysis is performed on personal credit data of the spending credit applicant. In the present embodiment, a principal component analysis is performed by using IBM SPSS Statistics 22 tool according to the personal credit data principal component analysis workflow, and the analysis result is shown in table 3.
According to the explanation of the total variance, the system default variance is greater than 1, so that only the first two main components are taken, the sum of the first two main components accounts for 85.235% of the total variance, the variance of the first main component is 10.477, and the variance of the second main component is 1.456.
TABLE 3 personal Credit data principal Components analysis results
The principal component loading matrix is shown in table 4.
TABLE 4 principal component load matrix
From this, a first principal component function expression can be found:
Yi＝0.262X1+0.271X2+0.273X3+0.271X4+0.266X5+0.289X6+0.294X7+0.278X8+0.280X9+0.291X 10+0.169X11+0.273X12+0.280X13+0.216X14
second principal component function expression:
Y2＝0.147X1-0.035X2+0.313X3-0.313X4+0.006X5+0.087X6+0.224X7-0.084X8-0.226X9+0.146X 10+0.616X11-0.065X12-0.226X13-0.470X14
from the first and second principal component functional expressions and the collection of the spending credit applicant personal credit data from the various data sources, the reduced-dimension spending credit applicant personal credit data may be obtained as shown in table 5.
TABLE 5 Credit applicant personal Credit data Table after dimensionality reduction
The fourth step: and constructing a self-organizing mapping neural network to generate a personal credit evaluation model. In the embodiment, the personal credit data of the consumption credit applicant after dimension reduction is used as a training data set, and a self-organizing mapping neural network is created by using a MATLAB R2016b tool.
The default value of the structure function is 'hextop', the default value of the distance function is 'linkdist', the default value of the learning rate of the classification stage is 0.9, the default value of the learning step length of the classification stage is 1000, the default value of the learning rate of the tuning stage is 0.02, and the default value of the neighborhood distance of the tuning stage is 1. Setting the competition layer to be 6 × 6 hexagonal structure and training times to be 300, the network topology structure can be obtained as shown in fig. 5. The direct distance situation of adjacent neurons is shown in fig. 6. The classification of each neuron is shown in fig. 7.
The fifth step: in order to verify the personal credit evaluation effect of the constructed self-organizing map neural network, the personal credit data of the consumer credit applicant U1 is used as the evaluation model to simulate the input data, so that the trained self-organizing map neural network can accurately classify the input samples, and the verification result is shown in FIG. 8.
And a sixth step: the system outputs the personal credit evaluation result of the consumption credit applicant to the commercial bank in the form of a personal credit evaluation report, and the commercial bank makes a decision whether the consumption credit applicant credits or not with the assistance of the personal credit evaluation report.
The seventh step: the system stores the personal credit evaluation result into the database in time, and provides accumulated data for subsequent personal credit evaluation as a personal credit data source, so that the accuracy of personal credit evaluation is improved.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (7)
1. A personal credit assessment method based on self-organizing map network, characterized in that the personal credit assessment method comprises the following steps:
s1, constructing a personal credit index system: according to the purpose and the requirement of personal credit evaluation, acquiring an index which can most reflect the personal credit characteristics of an evaluation object, and constructing and obtaining a personal credit index system; the personal credit evaluation index at least comprises personal natural characteristics, professional characteristics and bank relation;
s2, collecting personal credit data: acquiring personal credit data of a plurality of consumption credit applicants from different data sources according to a personal credit index system to generate a personal credit data sample;
s3, feature dimension reduction: performing principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of an evaluation object, and obtaining a personal credit data sample after dimension reduction;
s4, constructing a self-organizing mapping neural network, generating a personal credit evaluation model, importing the personal credit data sample after dimensionality reduction into the personal credit evaluation model, and training the personal credit evaluation model; the personal credit evaluation model comprises a processing unit array, a comparison selection unit, a local interconnection unit and an adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptive unit modifies the parameters of the excited processing unit to increase its output value corresponding to the specific input discriminant function;
s5, personal credit assessment: collecting personal credit data of an object to be evaluated, replacing original variables with combined comprehensive variables according to the principal component analysis result of the personal credit data, and inputting the combined comprehensive variables into a self-organizing mapping neural network for personal credit evaluation;
and S6, outputting an evaluation result: and outputting the personal credit evaluation result of the self-organizing mapping neural network to a commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
2. The personal credit evaluation method based on the self-organizing map network as claimed in claim 1, wherein in step S1, the indexes which are most capable of embodying the personal credit characteristics of the evaluation object include age, gender, marital status, health status, culture degree, unit property, industry status, position age, average family income, whether the employee is present, the account of the person, business transaction and borrowing status.
3. The personal credit evaluation method based on self-organizing map network of claim 1, wherein in step S3, IBM SPSS Statistics 22 tool is used to perform principal component analysis on personal credit data.
4. The personal credit evaluation method based on self-organizing map network as claimed in claim 1, wherein in step S3, the feature dimension reduction process comprises the following steps:
s31, feature centering: standardizing the imported personal credit data to obtain a standardized data matrix;
s32, calculating a covariance matrix: establishing a corresponding covariance matrix according to the standardized data matrix, wherein the covariance matrix is used for reflecting statistical indexes of the degree of closeness of correlation among the standardized data, and the larger the value is, the more necessary principal component analysis is carried out on the data;
s33, calculating characteristic values and characteristic vectors: solving the eigenvalue, the principal component contribution rate and the accumulated variance contribution rate according to the covariance matrix, and determining the number of the principal components;
s34, selecting a characteristic vector corresponding to the large characteristic value: sorting the eigenvalues in a descending order, and selecting eigenvectors corresponding to the largest eigenvalues as column vectors to form an eigenvector matrix;
s35, establishing an initial factor load matrix, and explaining principal components: the factor load is a correlation coefficient R (Zi, Xi) of the main component Zi and the original index Xi, and is used for expressing the degree of correlation between the main component Zi and each financial ratio;
and S36, solving the function expression of each principal component, and obtaining the personal credit data of the consumption credit applicant after dimensionality reduction according to the function expression of each principal component and the personal credit data of the consumption credit applicant acquired from a plurality of data sources.
5. The personal credit evaluation method based on self-organizing map network as claimed in claim 1, wherein in step S4, the construction process of the personal credit evaluation model comprises the following sub-steps:
s41, creating a self-organizing map neural network by using a MATLAB R2016b tool, and initializing the network; the default value of the structure function is 'hextop', the default value of the distance function is 'linkdist', the default value of the learning rate of the classification stage is 0.9, the default value of the learning step length of the classification stage is 1000, the default value of the learning rate of the tuning stage is 0.02, and the default value of the neighborhood distance of the tuning stage is 1;
s42, setting the competition layer to be a 6 multiplied by 6 hexagonal structure and training times to be 300;
s43, inputting personal credit data samples after dimension reduction;
s44, searching network winning nodes;
s45, defining a winning neighborhood;
s46, adjusting the network weight;
s47, inputting new sample data and repeating the above learning process until the learning rate decays to 0 or a predetermined positive fraction.
6. A personal credit evaluation system based on an ad hoc mapping network, the personal credit evaluation system comprising:
the personal credit index construction unit is used for acquiring indexes which can best reflect the personal credit characteristics of an evaluation object according to the purpose and the requirement of personal credit evaluation, and constructing and obtaining a personal credit index system; the personal credit evaluation index at least comprises personal natural characteristics, professional characteristics and bank relation;
the personal credit data acquisition unit is used for acquiring personal credit data of the consumption credit applicant from different data sources through a personal credit data acquisition agent according to a personal credit index system;
the personal credit data dimension reduction unit is used for carrying out principal component analysis on the personal credit data, reducing the personal credit characteristic dimension of an evaluation object and obtaining the personal credit data after dimension reduction;
the personal credit evaluation model is constructed on the basis of a self-organizing mapping neural network and comprises a processing unit array, a comparison selection unit, a local interconnection unit and a self-adaptive unit; the processing unit array comprises a plurality of processing units, each processing unit receives event inputs and forms a discriminant function for the signals; the comparison selection unit compares the discrimination functions of all the processing units and selects one processing unit with the maximum function output value; the local interconnection unit simultaneously excites the selected processing unit and the nearest processing unit; the adaptive unit modifies the parameters of the excited processing unit to increase its output value corresponding to the specific input discriminant function;
the model training unit is used for repeatedly inputting all personal credit data into the self-organizing mapping neural network for training to enable the weight value to tend to be stable, and the trained self-organizing mapping neural network is generated to serve as a personal credit evaluation model;
the personal credit evaluation unit is used for calling the personal credit data dimension reduction unit to perform principal component analysis results on the acquired personal credit data of the object to be evaluated, replacing original variables with combined comprehensive variables, and inputting the combined comprehensive variables into a personal credit evaluation model to perform personal credit evaluation;
and the personal credit evaluation result output unit is used for outputting the personal credit evaluation result output by the personal credit evaluation model to the commercial bank in the form of a personal credit evaluation report to assist the commercial bank in making a loan decision.
7. The personal credit evaluation system based on self-organizing map network as claimed in claim 6, wherein the personal credit evaluation system further comprises a personal credit database unit for storing personal credit evaluation results, which is responsible for daily data maintenance and management by the authority.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111683268.6A CN114298834A (en) | 2021-12-31 | 2021-12-31 | Personal credit evaluation method and system based on self-organizing mapping network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111683268.6A CN114298834A (en) | 2021-12-31 | 2021-12-31 | Personal credit evaluation method and system based on self-organizing mapping network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114298834A true CN114298834A (en) | 2022-04-08 |
Family
ID=80975510
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111683268.6A Withdrawn CN114298834A (en) | 2021-12-31 | 2021-12-31 | Personal credit evaluation method and system based on self-organizing mapping network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114298834A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117057852A (en) * | 2023-10-09 | 2023-11-14 | 北京光尘环保科技股份有限公司 | Internet marketing system and method based on artificial intelligence technology |
| CN117455555A (en) * | 2023-12-25 | 2024-01-26 | 厦门理工学院 | Big data-based electric business portrait analysis method and system |
| CN117670525A (en) * | 2023-12-22 | 2024-03-08 | 广东金融学院 | Enterprise credit assessment system based on big data analysis |
-
2021
- 2021-12-31 CN CN202111683268.6A patent/CN114298834A/en not_active Withdrawn
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117057852A (en) * | 2023-10-09 | 2023-11-14 | 北京光尘环保科技股份有限公司 | Internet marketing system and method based on artificial intelligence technology |
| CN117057852B (en) * | 2023-10-09 | 2024-01-26 | 头流(杭州)网络科技有限公司 | Internet marketing system and method based on artificial intelligence technology |
| CN117670525A (en) * | 2023-12-22 | 2024-03-08 | 广东金融学院 | Enterprise credit assessment system based on big data analysis |
| CN117455555A (en) * | 2023-12-25 | 2024-01-26 | 厦门理工学院 | Big data-based electric business portrait analysis method and system |
| CN117455555B (en) * | 2023-12-25 | 2024-03-08 | 厦门理工学院 | Big data-based electric business portrait analysis method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108231201B (en) | Construction method, system and application method of disease data analysis processing model | |
| Tang et al. | A pruning neural network model in credit classification analysis | |
| CN110956273A (en) | Credit scoring method and system integrating multiple machine learning models | |
| CN114298834A (en) | Personal credit evaluation method and system based on self-organizing mapping network | |
| CN111785329B (en) | Single-cell RNA sequencing clustering method based on countermeasure automatic encoder | |
| CN107067182A (en) | Towards the product design scheme appraisal procedure of multidimensional image | |
| CN110837523A (en) | High-confidence reconstruction quality and false-transient-reduction quantitative evaluation method based on cascade neural network | |
| CN115688024B (en) | Network abnormal user prediction method based on user content characteristics and behavior characteristics | |
| CN110706095A (en) | Target node key information filling method and system based on associated network | |
| Chen et al. | Research on credit card default prediction based on k-means SMOTE and BP neural network | |
| CN115205011B (en) | Bank user portrait model generation method based on LSF-FC algorithm | |
| Peng et al. | The health care fraud detection using the pharmacopoeia spectrum tree and neural network analytic contribution hierarchy process | |
| CN112949954A (en) | Method for establishing financial fraud recognition model based on recognition learning | |
| Chen et al. | A deep learning approach using DeepGBM for credit assessment | |
| Zhang et al. | Research on borrower's credit classification of P2P network loan based on LightGBM algorithm | |
| Zhu et al. | Loan default prediction based on convolutional neural network and LightGBM | |
| CN111506813A (en) | Remote sensing information accurate recommendation method based on user portrait | |
| Wu et al. | Customer churn prediction for commercial banks using customer-value-weighted machine learning models | |
| CN114529063A (en) | Financial field data prediction method, device and medium based on machine learning | |
| CN113177733B (en) | Middle and small micro enterprise data modeling method and system based on convolutional neural network | |
| CN114970684A (en) | Community detection method for extracting network core structure by combining VAE | |
| Li et al. | University Students' behavior characteristics analysis and prediction method based on combined data mining model | |
| Kulothungan | Loan Forecast by Using Machine Learning | |
| CN113408908A (en) | Multi-dimensional credit evaluation model construction method based on performance ability and behaviors | |
| Setnes et al. | Fuzzy target selection in direct marketing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20220408 |
|
| WW01 | Invention patent application withdrawn after publication |